On 09/15/08 18:36, Seth Goldberg wrote: > >> We also have to investigate how expensive it is to reprogram >> the I/O APIC to target the next waking CPU every time. > > Yuck -- because of previous IOAPIC errata, I'm really not happy > about touching the IOAPICs once they've been programmed. > > --S
That is both bad and good news. ;-) Having the I/O APIC target HPET interrupts to one CPU does not seem to make a noticeable difference in that CPU's load. HPET interrupt firing frequency goes down as system load goes up. Reprogram the I/O APIC to target the next waking CPU probably will not buy us any noticeable performance. Thank you, Bill
